A product configurator has been an effective software tool in successful implementation of mass customization strategy. It enables manufacturers to automatically generate product configuration information tailored to individual customers’ needs. However, current product configurator techniques are not adequate to solve an engineering product configuration problem, because the constraints in such a problem are often expressed by mathematical formulae and computable procedures. This type of constraint posts challenges on constraint modelling and solving within the constraint satisfaction paradigm. Constraints made up of mathematical formulae and computable procedures are not naturally supported with pre-defined constraint semantics in a constraint model. It is also difficult to achieve search efficiency for constraints over n-ary continuous variables. This paper presents an innovative approach to modelling and solving an engineering product configuration problem based on the constraint satisfaction paradigm. It aims at developing a methodology for a generic configurator that is able to solve an engineering product configuration problem with complex constraints.